You should always test multiple versions of an ad. I usually create at least six ads (mix up the text, try different phrases, and so on) and then let it run for a few thousand impressions or two weeks (depends on the item). Then you look at the ad CTR and delete the weak ads. Take the strong ads and create more variations on those. This lets you discover the best ads.
Ideally, you should test a different set of ads for each ad group or group of keywords. At least slightly different. For example, at least try different headlines from Ad Group to Ad Group.
Although a winning ad from one ad group may also be a winner when run in other ad groups, that's not always the case because different keywords and different keyword groups tend to reflect different needs and wants of the market, if only in a small way.
You should also run multiple ads within an adgroup. You have two choices, leave "optimization" on and G will select the best ad, or for better results, turn optimization off, then after a week or so run a keyword report, segmented by ad, and see how each ad performs on an individual keyword. You can then pick the best ad for each keyword. We do this with our software product we are developing and the resulting increase in response is amazing. Feel free to contact me if you would like more details.